Cereblon

Cereblon (CRBN) was originally assigned a role in memory and learning, because a nonsense mutation, leading to an abnormal C-terminal truncation of the protein, was discovered as the cause of an autosomal recessive nonsyndromic form of mental retardation (Higgins, 2004; Higgins, 2000). This role of CRBN was validated in a forebrain-specific conditional and full CRBN^-/- knockout mouse model (Rajadhyaksha, 2012). CRBN was further characterized to have a ubiquitous subcellular localization pattern, with enrichment in the perinuclear region and to be expressed in various tissues (Jo, 2005; Xin, 2008). It contains a Lon domain, important for oligomerization of adenosine triphosphate (ATP)-dependent proteases and chaperones, a regulator of G-protein signaling (RGS) domain and a leucine zipper motif important for protein-protein interaction (Jo, 2005; C. K. Smith, 1999).

Furthermore, CRBN associates with and regulates the function of voltage-gated ion-channels (Hohberger, 2009; Jo, 2005).

1.3.1 CRBN and IMiDs

In their landmark study, Ito and colleagues identified CRBN as the cellular target of thalidomide. Using thalidomide-conjugated beads, they searched for thalidomide-interacting proteins in whole cell extracts of HeLa cells and found CRBN and damaged DNA binding protein 1 (DDB1) in mass-spectrometric analyses (Ito, 2010). DDB1, cullin 4 (CUL4), the regulator of cullins 1 (ROC1) and a substrate receptor, usually form a cullin4-RING E3 ubiquitin ligase (CRL4), a multisubunit protein complex shown to polyubiquitinate proteins meant for proteasomal degradation (Angers, 2006). They proposed CRBN to function as the substrate receptor, or DDB1-CUL4-associated factor (DCAF), and indeed could show ubiquitin ligase activity of the complex and found it to be inhibited by thalidomide.

Furthermore, they demonstrated that a knockdown of CRBN specifically phenocopies the teratogenic effects of thalidomide in zebrafish- and chicken-models by modulating fibroblast growth factor 8 (fgf8) levels. They therefore postulated that thalidomide exerts its teratogenic activity by binding to CRBN and inhibiting its

ubiquitin ligase activity (Ito, 2010). The next obvious question, whether CRBN is also responsible for thalidomide’s actions in MM, was answered by Zhu and colleagues.

Using flow cytometry and proliferation assays, they unveiled that knockdown of CRBN in MM cell lines is cytotoxic and leads to a decrease in proliferation. The loss of CRBN in myeloma cells also leads to lenalidomide- and pomalidomide-specific resistance and vice versa, acquired IMiD-resistance is associated with low CRBN levels. In addition, lenalidomide-mediated changes in gene expression are substantially reduced in CRBN-depleted cells. Their findings suggest that CRBN is necessary for lenalidomide’s anti-myeloma actions and provide evidence for a common pathway of the teratogenic and anti-myeloma effects of IMiDs (Zhu, 2011).

Another group further succeeded to clarify the interaction of IMiDs and CRBN (Lopez-Girona, 2011; Lopez-Girona, 2012): Namely, thalidomide’s inhibitory effect on CRBN autoubiquitylation extends to both lenalidomide and pomalidomide. Moreover, depletion of CRBN abrogates lenalidomide’s and pomalidomide’s anti-proliferative potency in MM cell lines. Lenalidomide and pomalidomide also downregulate interferon-regulatory factor 4 (IRF4), an established MM cell survival factor and induce the cell cycle inhibitory protein p21^WAF-1 in a CRBN-dependent manner (Lopez-Girona, 2011). Furthermore, CRBN is necessary for IMiD-induced activation of the cytokines interleukin-2 (IL-2) and TNF-α in T cells (Lopez-Girona, 2012).

These findings link CRBN both to the anti-proliferative and immunomodulatory activity of IMiDs.

1.3.2 IMiDs modulate CRBN ligase activity

The identification of CRBN as the key cellular IMiD-interacting protein clearly improved the understanding of IMiD biology. However, exactly how CRBN mediates the effect of IMiDs remained unclear. In an attempt to identify downstream interactors/substrates of CRBN, four groups simultaneously performed luciferase-assay and mass spectrometry-based analyses of cells with or without IMiD treatment. They identified two zinc finger transcription factors of the Ikaros family, Ikaros (IKZF1) and Aiolos (IKZF3), to be affected by IMiD treatment (A. K. Gandhi, 2014a; Kronke, 2014; Lu, 2014; Zhu, 2014). IKZF1 and IKZF3 are lymphoid transcription factors that have been implicated in various stages of B- and T-cell differentiation (Cortes, 1999). Particularly IKZF3 has been shown to be crucial in the development of long-lived high-affinity plasma cells in the bone marrow (Cortes, 2004). IKZF1 and IKZF3 protein levels are downregulated upon IMiD treatment of MM cell lines and patient-derived primary MM cells, while mRNA-levels remain stable. Depletion of IKZF1 and IKZF3 leads to a downregulation of IRF4 and decreases the proliferation of IMiD-sensitive MM cell lines. This links IKZF1 and IKZF3 downregulation to the anti-myeloma activity of IMiDs (Kronke, 2014; Lu, 2014). It has been previously demonstrated that IKZF1 and IKZF3 are negative regulators of IL-2 expression (R. Gandhi, 2010). Gandhi and colleagues observed

that IMiD-induced IL-2 secretion is mimicked by the depletion of IKZF1 or IKZF3 in T-cells. They therefore conclude that lenalidomide- and pomalidomide-induced IL-2

elevation in T-cells is mediated by CRBN-dependent degradation of the IKZF1 and IKZF3 transcription factors and the resulting de-repression of IL-2 transcription. The same group also assessed IKZF1 and IKZF3 expression levels in peripheral blood-derived T-cells of healthy volunteers before and after lenalidomide treatment and observed a decrease in IKZF3 levels upon treatment. They therefore propose IKZF3 as a new biomarker for lenalidomide activity in MM (A. K. Gandhi, 2014a).

On the mechanistic side, IKZF1 and IKZF3 were shown to interact with CRBN within the CRL4^CRBN ubiquitin ligase complex. This interaction, however, only occurs upon IMiD treatment. While previous studies suggested IMiDs to inhibit CRBN ubiquitin ligase activity, Ghandi, Lu, Krönke, Zhu and colleagues have shown that lenalidomide reprograms the CRBN ligase complex to bind to its substrates IKZF1 and IKZF3, leading to their polyubiquitynation and subsequent proteasomal degradation. These findings suggest an attractive model, in which a ubiquitin ligase complex such as CRL4^CRBN can be steered by small molecules to target specific proteins. Further, the authors assumed that teratogenicity and IMiD activity in the non-lymphoid context might be mediated by different CRBN substrates (Kronke, 2014; Lu, 2014). Indeed, in a follow-up study in del(5q) MDS, Krönke and colleagues have shown lenalidomide-induced degradation of casein kinase 1A1 (CK1α) to be responsible for IMiD activity in this hematologic entity (Kronke, 2015). Crystal structure analyses of CRBN with thalidomide, lenalidomide and pomalidomide have revealed that IMiDs bind to a hydrophobic pocket in the C-terminal region of CRBN with their common glutarimide ring. The phthalimide moiety, which varies among IMiDs (Figure 1), is exposed on the surface of the CRBN protein, creating an interface of unsatisfied bonding potential for new interactions. Of note, the hydrophobic binding pocket is highly conserved and therefore physiologic IMiD-competing endogenous ligands are very likely (Chamberlain, 2014; Fischer, 2014). In Figure 2: IMiDs modulate the CRL4^CRBN complex to ubiquitinate IKZF1/3. This leads to protein degradation in the proteasome and is responsible for some anti-myeloma effects of IMiDs. Adapted from (Stewart, 2014)

an elegant approach, a group at the Dana-Farber Cancer Institute combined a thalidomide analogue with a small molecule inhibitor of bromodomain-containing protein 4 (BRD4), which is involved in MYC signaling, and demonstrated that CRL4^CRBN selectively degrades BRD4 upon exposure to this thalidomide-conjugated inhibitor. They infer that many previously intractable proteins might be selectively degradable by using phthalimide-conjugated ligands with or without intrinsic inhibitory activity (Winter, 2015).

The IMiD-induced selective degradation of IKZF1, IKZF3 or CK1α by an E3 ubiquitin ligase complex involving CRBN explains some of the anti-proliferative and immunomodulatory effects in MM and del(5q) MDS. Nevertheless, this theory lacks explanations for the anti-angiogenic and teratogenic potential of IMiDs. Furthermore, it creates a paradox, because in patients, proteasome inhibitors like bortezomib, carfilzomib and ixazomib show synergistic activity with IMiDs (S. K. Kumar, 2014a;

Richardson, 2010; Roussel, 2014; Stewart, 2015). In theory, however, proteasome inhibition would antagonize the IMiD induced degradation of IKZF1, IKZF3 and CK1α. A ubiquitin-independent physiologic function of CRBN might be able to explain these unanswered questions.

1.3.3 CRBN exerts a chaperone-like function for the CD147/MCT1 complex

In their search for physiological CBRN interactors, my colleagues in the lab of Prof. Bassermann performed an affinity purification of CRBN, followed by mass spectrometry and identified cluster of differentiation 147 (CD147), which is also known as basigin or extracellular matrix metalloproteinase inducer (EMMPRIN), and monocarboxylate transporter 1 (MCT1) as specific interactors. Functions of these proteins are investigated in 1.4. They found that the interaction between CRBN and these two membrane proteins, in contrast to IKZF1 and IKZF3, is lost upon lenalidomide exposure (Eichner, 2016). Moreover, knockdown of CRBN leads to CD147 and MCT1 destabilization, while forced expression stabilizes both proteins (Eichner, 2016). This suggests that CRBN exerts a stabilizing, e.g. chaperone-like function on CD147 and MCT1, which compete with IMiDs for CRBN-binding. Indeed, CRBN interacts with freshly synthetized CD147 and MCT1, while lenalidomide treatment or CRBN depletion result in accumulation of the membrane proteins at the

Figure 3: CRBN exerts a chaperone-like function for the CD147/MCT1 complex. a)

thalidomide lenalidomide pomalidomide

MCT1 CD147

MCT1CD147

a b

endoplasmatic reticulum (Eichner, 2016). Their size exclusion chromatography experiments of CRBN immunoprecipitates showed that CRBN exists in two different complexes, one containing the usual components of the CRL4^CRBN ligase and another with CD147 and MCT1. In addition, immunoprecipitation of CUL4A failed to retrieve both CD147 and MCT1. Taken together, these findings imply that CRBN exerts a ubiquitin-independent chaperone-like function on CD147 and MCT1 and plays an important role in the maturation of these two membrane proteins. The functional relevance of this mechanism in MM is presented in the results section of this thesis. Apart from the relevance in MM, further studies from our group have linked CD147 to IMiD-induced teratogenicity in a zebrafish model. The morpholino-induced specific knockdown of CD147 phenocopies the teratogenic effects of thalidomide, resulting in a dose-dependent reduction of fin-, head- and eye-size (Eichner, 2016). Likewise, CD147 depletion reduces the expression of fgf8 in pectoral fin buds, an effect also observed in thalidomide-treated zebrafish (Ito, 2010).

This effect is thalidomide-specific and does not extend to lenalidomide or pomalidomide, which is in line with the fact that in zebrafish, CD147 is only destabilized by thalidomide (Eichner, 2016).

1.4 CD147 & MCT1

1.4.1 CD147 and its role in malignant diseases

CD147 is a transmembrane glycoprotein related to the immunoglobulin superfamily of receptors. It exists in various species and is also known as EMMPRIN or basigin. It is made up of 269 amino acids and can be divided into an extracellular domain, which contains two

immunoglobulin-like structures with three glycosylation sites, a short highly-conserved hydrophobic transmembrane-region and a 39 amino acid C-terminal intracellular domain. The molecular weight of CD147 varies from 29-65kDa, depending on the glycosylation pattern. The low molecular weight core-glycosylated protein is the

immature form, while the highly glycosylated form of CD147 is considered to be the active form. CD147 has been shown to interact with several proteins including integrins, cyclophilin-A, caveolin-1 and two members of the moncarboxylate transporter family MCT1 and MCT4 (Iacono, 2007). The name extracellular matrix metalloproteinase inducer (EMMPRIN) derives from the observations that CD147 mediates tumor invasion, growth, progression and metastasis by inducing MMP

CD147 protein. Adapted from (Xiong, 2014)

production (Xiong, 2014). In addition, several studies could demonstrate that CD147 plays a crucial role in angiogenesis by promoting the secretion of VEGF both by tumor cells directly and by inducing VEGF secretion in the microenvironment (Bougatef, 2009; Y. Chen, 2012; Tang, 2005). CD147 is expressed in various tissues, including actively proliferating and differentiating epithelial, myocardial, vascular endothelial cells of the brain, most cells of the hematopoietic system and in almost all types of cancer tissue (Riethdorf, 2006). It is overexpressed and described as a marker of poor prognosis in some tumor entities, such as breast cancer, serous ovarian and bladder carcinoma (Weidle, 2010). CD147 has also been implicated in multidrug resistance in different types of cancer (Weidle, 2010). CD147-directed monoclonal antibodies are currently being evaluated pre-clinically in hepatocellular carcinoma and head and neck squamous cell carcinoma and show promising results regarding prevention of metastasis, invasion and angiogenesis (Xiong, 2014).

1.4.2 Cancer cell metabolism and the Warburg Effect

In regular cellular metabolism, under normoxic conditions, cells take up glucose and process it to pyruvate by glycolysis. Pyruvate is then passed on to the mitochondrial citric acid cycle and oxidative phosphorylation resulting in carbon dioxide, water and 36 molecules of ATP per processed molecule of glucose. Under anaerobic or hypoxic conditions, cells neglect oxidative phosphorylation and instead upregulate the less efficient anaerobic glycolytic pathway, which produces lactate and 2 molecules of ATP (Berg, 2002). In the 1920s, Otto Warburg, while comparing metabolic respiratory rates of tumor tissues with those of normal liver and kidney tissues, observed that cancer cells with functioning mitochondria retain a glycolytic metabolic pattern even under normoxic conditions. This phenomenon has been named “aerobic glycolysis” or the Warburg Effect (Liberti, 2016; Warburg, 1925).

Aerobic glycolysis has been shown to be associated with hypoxia-independent activation of hypoxia-inducible factors (HIF) by well-known oncogenes like RAS, MYC and mutated tumor suppressors (Hanahan, 2011). The reasons for the glycolytic switch are still not fully understood, there are, however, several theories.

One reason may be the rapid generation of ATP by glycolysis. The rate of lactate-producing glycolysis is 10-100 times faster compared to oxidative phosphorylation and therefore the net ATP obtained by both pathways is almost equal (Liberti, 2016).

Another theory states that aerobic glycolysis might aid cancer cells by increasing glucose uptake and synthesis of amino acids, nucleotides and lipids, which are urgently needed in cells with uncontrolled proliferation. In addition to ATP, the biosynthesis of such macromolecules relies on the reducing equivalents NADH and NADPH, which are generated as a byproduct of glycolytic metabolism (Vander Heiden, 2009). Next, the tumor microenvironment is influenced by elevated lactate levels and decreased extracellular pH. Reports have shown that H⁺ ions diffuse into neighboring healthy tissue causing tissue remodeling, which ultimately favors

invasion and metastasis (Estrella, 2013). Finally, the Warburg Effect has been proposed to directly affect signal transduction in tumor cells, by generating and modulating reactive oxygen species and altering chromatin structure and the epigenetic pattern of certain growth genes (Liberti, 2016). The complex deregulation of normal energy metabolism is considered to be one of the new emerging hallmarks of cancer (Hanahan, 2011). Indeed, a recent study has demonstrated that MM cells also depend on aerobic glycolysis and produce significantly higher amounts of lactate compared to normal blood mononuclear cells under normoxic conditions (Sanchez, 2013). The interruption of the glycolytic pathway by dichloroacetate induces apoptosis, superoxide production, decreases proliferation and increases sensitivity to proteasome inhibitors like bortezomib in MM cell lines (Sanchez, 2013).

In addition, pyruvate dehydrogenase kinase 1 (PDK1), which inhibits pyruvate dehydrogenase (PDH), the gatekeeping enzyme limiting the conversion of pyruvate to acetyl-CoA used in the citric acid cycle, and other glycolytic enzymes such as glucose transporter 1 (GLUT1) and lactate dehydrogenase A (LDHA) are overexpressed in MM patient samples (Fujiwara, 2013).

1.4.3 Hypoxia, the bone marrow and MM

Hypoxia is a state of imbalance between oxygen consumption and availability. It is a common feature of the tumor microenvironment and has been implicated in disease progression and treatment resistance. Hypoxia in solid tumors can be explained by massive cell proliferation and concurrent shortage of perfusion by pre-existing blood vessels or frequently aberrant and insufficient neo-vascularization. Therefore, hypoxic tumor cells rely on oxygen diffusion and acquire more aggressive and drug resistant phenotypes by upregulating hypoxia inducible factor-1α (HIF-1α) (Martin, 2011). Notably, the bone marrow microenvironment is characterized by local hypoxia. This has been established by invasive studies in orthopedic and MM patients (Colla, 2010; Watanabe, 2007). Studies conducted on murine bone marrow using markers for hypoxia have shown that oxygen tension varies within the bone marrow. Lowest oxygen tensions were observed in compartments close to the edge of the bone marrow, which coincide with a high abundance of hematopoietic stem cells (Martin, 2011; Parmar, 2007). In MM cells, HIF-1α is upregulated both due to local hypoxia and oxygen-independent aberrant signaling. HIF-1α contributes to MM pathogenesis by deregulating energy metabolism and inducing anti-apoptotic proteins as well as the secretion of angiogenic factors like VEGF (Borsi, 2015). Specific inhibition of HIF-1α causes reduced viability in MM cell lines via cell cycle arrest and apoptosis (Borsi, 2014).

TH-302, a prodrug of a cytotoxin, is only activated under hypoxic conditions and is currently being studied in phase I and II clinical trials in combination with dexamethasone alone or with bortezomib for patients with relapsed or refractory MM

(Borsi, 2015; Hu, 2010). To conclude, selective targeting of MM via the hypoxic bone marrow niche seems like a very promising therapeutical approach.

1.4.4 MCT1 and its role in malignant diseases

Until the 1970s,

by α-cyano-4-hydroxycinnamate (CHC) and therefore proposed the involvement of transporters (Halestrap, 1974). The solute carrier 16 (SLC16) or monocarboxylate transporter (MCT) family consists of 14 members and is characterized by 12 transmembrane domains, an N- and C-terminal intracellular tail and an intracellular loop between the transmembrane domains 6 and 7 (Halestrap, 2004). MCTs catalyze the symport of monocarboxylate anions together with a proton across membranes, following a concentration gradient in a process that does not require ATP. They also mediate the exchange of intra- and extracellular monocarboxylates (Halestrap, 2013). MCT1, encoded by the SLC16A1 gene, is ubiquitously expressed and is the most characterized member of the MCT family. MCT1 can transport monocarboxylates equally well across membranes in both directions. Due to its high abundance and its leading role in anaerobic and aerobic glycolysis, L-lactate is by far the most relevant substrate (Halestrap, 2013). MCT1-facilitated lactate and proton efflux across the plasma membrane contributes to the regulation of intracellular pH and therefore is of vital importance to cell survival during periods of hypoxia or ischemia. Some tissues such as white muscle and some invasive tumors preferentially use MCT4, while MCT1 serves as a lactate importer in red skeletal muscle cells, supplying these cells with lactate for oxidative metabolism (Halestrap, 2012). MCT1 is overexpressed in various solid tumors including lung cancer, colorectal carcinoma, pancreatic carcinoma, glioma, neuroblastoma and melanoma (Kennedy, 2010). Because lactate efflux is important in tumor cell metabolism, MCTs have recently been investigated as potential targets in anti-cancer therapy.

Knockdown of MCT1 via siRNA or treatment with the MCT1/2-specific inhibitor AR-C155858 increases intracellular lactate levels, reduces rates of glycolysis and displays anti-proliferative and cytotoxic effects in fibroblasts and a xenograft mouse model (Le Floch, 2011). Similar results have been achieved in small cell lung cancer cell lines in normoxia and hypoxia with the MCT1-specific inhibitor AZD3965, which

extracellular

is currently being evaluated in a phase I clinical trial for patients with advanced solid tumors in the United Kingdom (Polanski, 2014).

1.4.5 CD147, MCT1 and MM

CD147 and MCT1 have been shown to interact strongly and form a transmembrane complex, most likely through the transmembrane and intracellular domain of CD147 (Kirk, 2000). Co-transfection experiments have shown that correct co-localization at the cell membrane occurs only upon interaction of CD147 and MCT1, suggesting that CD147 acts as a chaperone for MCT1 and thereby facilitates lactate transport (Kirk, 2000; Le Floch, 2011). Likewise, proper maturation of CD147 partly depends on MCT1 expression, as a knockdown of MCT1 leads to an accumulation of the immature core-glycosylated form of CD147. This implies that CD147 and MCT1 regulate each other by acting as co-chaperones to form a transmembrane complex (Deora, 2005).

Both CD147 and MCT1 are implicated in MM pathogenesis. CD147 is elevated in MM cells compared to MGUS and normal B cells, on both mRNA and protein level. Progression of disease correlates with increased CD147 cell surface expression levels (Arendt, 2012). The natural CD147 ligand and activator cyclophillin B increases proliferation of MM cells in a CD147-dependent manner (Arendt, 2012).

High CD147 cell surface levels are associated with rapid proliferation, whereas downregulation of CD147 decreases proliferation (Arendt, 2012). Microvesicles are thought to mediate intercellular communication and those released by MM cells show high CD147 levels. These microvesicles with high CD147 levels are able to stimulate MM cell proliferation, while microvesicles derived from CD147-silenced cells fail to do so (Arendt, 2014). Recently, CD147 has been shown to be involved in MM cell

High CD147 cell surface levels are associated with rapid proliferation, whereas downregulation of CD147 decreases proliferation (Arendt, 2012). Microvesicles are thought to mediate intercellular communication and those released by MM cells show high CD147 levels. These microvesicles with high CD147 levels are able to stimulate MM cell proliferation, while microvesicles derived from CD147-silenced cells fail to do so (Arendt, 2014). Recently, CD147 has been shown to be involved in MM cell

Im Dokument IMiDs mediate their anti-myeloma activity via destabilization of the CD147/MCT1 complex  (Seite 15-25)